Streams of flowing data resolve into 0 and 1 integers. Credit: Shutterstock.
The grant will fund work to address key challenges in machine learning and Big Data analytics. Image credit: Shutterstock.

Two million Euro grant awarded for fundamental research in machine learning

Associate Professor Patrick Rebeschini, from the University of Oxford’s Department of Statistics, has been awarded a 2022 European Research Council (ERC) Consolidator Grant, worth €2 million. Working at the intersection of statistics and computer science, Professor Rebeschini will use the grant to develop theoretical foundations to advance machine learning methods to analyse Big Data.

Associate Professor Patrick Rebeschini. Image credit: Mel Cunningham.Associate Professor Patrick Rebeschini. Image credit: Mel Cunningham.
Professor Rebeschini said: ‘The ERC Consolidator Grant represents an exciting opportunity to pursue a multidisciplinary vision and uniquely position my research profile at the intersection of statistics, probability, and optimisation. I look forward to using it to achieve transformative impact, with downstream applications in many areas of machine learning.’

His aim is to use the grant to address a key challenge in machine learning: optimising statistical algorithms for Big Data so that these have minimal errors in accuracy but remain efficient to run.

As he explained: ‘In the era of Big Data—characterized by large, high-dimensional and distributed datasets—we are increasingly faced with the challenge of establishing scalable methods that can achieve optimal statistical guarantee under computational constraints. In machine learning, larger input datasets are preferred as these increase the accuracy of the resulting model that is trained on them. But typically, as the amount of data gets larger, this increases the time required for algorithms to run, and the computational power required.'

'My research focuses on developing a theoretical and methodological framework to design algorithms that can simultaneously be statistically and computationally "optimal”, that is, achieving the best statistical error that we can hope for with the lowest possible use of computational resources.’ 

As machine learning and artificial intelligence become ever more prevalent, there is an urgent need for new frameworks that can reduce the computational power these require, without sacrificing the accuracy of their outputs. Otherwise, as Professor Rebeschini warned, we face an ‘explosion’ in the amount of energy that these will consume, hindering efforts to achieve net zero carbon emissions.

ERC Consolidator Grants are awarded to ‘help excellent scientists, who have 7 to 12 years’ experience after their PhDs, to pursue their most promising ideas.’ Out of the 2,222 applicants this year, only 321 (14%) were successful.

There is a very strong machine learning community at Oxford, and I have found it very valuable to engage with colleagues across the University, including the Mathematical Institute, the Department of Computer Science and the Department of Engineering Science, besides the Department of Statistics. This interaction has helped to develop my ideas for this proposal.

Associate Professor Patrick Rebeschini, Department of Statistics, University of Oxford.

Professor Rebeschini said: ‘Winning this grant is a stepping stone for my career, and presents a great opportunity both for my research and development as a leader within my field. The award is particularly satisfying because my work is highly multidisciplinary, and it can be more challenging to attract funding for research that doesn’t sit within the “traditional” silos.’

President of the European Research Council Professor Maria Leptin said: ‘ERC Consolidator Grants support researchers at a crucial time of their careers, strengthening their independence, reinforcing their teams and helping them establish themselves as leaders in their fields. And this backing above all gives them a chance to pursue their scientific dreams.’

The holding of ERC awards by researchers based at UK institutions is subject to formalisation of the UK’s association to Horizon Europe, which remains the stated priority for the UK government, in line with the Trade and Cooperation Agreement agreed between the UK and the EU in December 2020. In the event that association is not confirmed by the final date for signature of grant agreements then the UK government’s Horizon Europe funding guarantee will apply, with UK awardees receiving equivalent funding via UKRI.